Crowdsourcing Twitter annotations to identify first-hand experiences of prescription drug use
详细信息 查看全文
- 作者：Nestor Alvaroa ; b ; nestoralvaro@nii.ac.jp" class="auth_mail" title="E-mail the corresponding author ; Mike Conwayc ; mike.conway@utah.edu" class="auth_mail" title="E-mail the corresponding author ; Son Doand ; Son.Doan@kp.org" class="auth_mail" title="E-mail the corresponding author ; Christoph Lofie ; lofi@ifis.cs.tu-bs.de" class="auth_mail" title="E-mail the corresponding author ; John Overingtonf ; john.overington@stratifiedmedical.com" class="auth_mail" title="E-mail the corresponding author ; Nigel Collierg ; h ; nhc30@cam.ac.uk" class="auth_mail" title="E-mail the corresponding author
- 关键词：Crowdsourcing ; Pharmacovigilance ; Twitter ; Natural language processing
- 刊名：Journal of Biomedical Informatics
- 出版年：2015
- 出版时间：December 2015
- 年：2015
- 卷：58
- 期：Complete
- 页码：280-287
- 全文大小：659 K
文摘
- •
Study on the automatic identification of first-hand drug intake reports in Twitter.
- •
Crowd-sourced judgements compared against expert judgements show moderate agreement.
- •
The experimental set up used 6 machine learning models with 7 different feature sets.
- •
Bayesian Generalized Linear Model performs the best (F1 = 0.64 and Informedness = 0.43).